Development of Fe-GAC Adsorbent for Arsenic Removal

Transcription

1 Technical Report No: ND12-06 Development of Fe-GAC Adsorbent for Arsenic Removal Qigang Chang Environmental and Conservation Science Graduate Program Wei Lin Dept. of Civil Engineering North Dakota State University Fargo, North Dakota June 2012 North Dakota Water Resources Research Institute North Dakota State University, Fargo, North Dakota

2 Technical Report No: ND12-06 Development of Fe-GAC Adsorbent for Arsenic Removal Qigang Chang WRRI Graduate Research Fellow Environmental and Conservation Sciences Graduate Program Wei Lin Associate Professor, Department of Civil Engineering North Dakota State University June 2012 The work upon which this report is based was supported in part by federal funds provided by the United States of Department of Interior in the form of ND WRRI Graduate Research Fellowship for the graduate student through the North Dakota Water Resources Research Institute. Contents of this report do not necessarily reflect the views and policies of the US Department of Interior, nor does mention of trade names or commercial products constitute their endorsement or recommendation for use by the US government. Project Period: March 1, February 28, 2011 Project Number: 2008ND168B and 2009ND186B North Dakota Water Resources Research Institute Director: G. Padmanabhan North Dakota State University Fargo, North Dakota 58108

3 TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES iv ACKNOWLEDGMENTS v ABSTRACT 1 INTRODUCTION 2 MATERIALS AND METHODS 3 Materials 3 Preparation of Fe-GAC 3 Characterization of Fe-GAC 3 Iron Desorption Test 4 Adsorption Kinetics Tests 3 Adsorption Isotherm Tests 4 Fe-GAC Surface Area and Pore Volume Measurements 5 Column Test 5 Analytical Methods 4 RESULTS AND DISCUSSION 6 Preparation of Fe-GAC 6 Stability of Impregnated Iron 6 Characterization of Fe-GAC 9 ph Impact on Arsenic Adsorption 12 Arsenic Adsorption Equilibrium Time 14 As(III) and As(V) Adsorption Isotherms 15 Adsorption Affinity of Arsenic on Fe-GACs 18 Impact of Iron Content of Fe-GACs on Arsenic Adsorption Capacities 19 BET Surface Area and Pore Structures of Fe-GACs 20 Application for Arsenic Removal from Groundwater 24 Effect of Iron Content on Arsenic Breakthrough 26 REFERENCES 29 ii

4 LIST OF TABLES Table 1. Dissociation Constants of Iron Hydroxides... 8 Table 2. Langmuir Model Parameters for As(III) and As(V) Adsorption on Fe-GACs Table 3. Re-estimated q m in Langmuir Model with same b Values for As(III) and As(V) Adsorption on Fe-GACs Table 4. Texture Parameters for Fe-GACs and Virgin GAC Table 5. Water Quality of Groundwater Table 6. Langmuir parameters for groundwater isotherm tests with constant b at L/µg.. 26 iii

5 LIST OF FIGURES Figure 1. Increment of iron content with repetitions in the synthesis process Figure 2. EDS spectra, a: Darco Fe-GAC (12.68%) without post-treatment, b: Darco Fe-GAC (9.29%) with post-treatment Figure 3. Stability of impregnated iron on Darco Fe-GAC with 4.22% iron Figure 4. Iron distribution on the cross-section of Darco Fe-GACs Figure 5. EDS analysis on the cross-section of Darco Fe-GAC with 4.22% iron Figure 6. Mapping of iron distribution on cross-section of Darco Fe-GAC with 9.29% iron. a:the cross-section of Fe-GAC; b: the mapping of iron Figure 7. SEM images of Norit RX3 EXTRA Fe-GAC with 8.52% iron Figure 8. SEM images of Darco Fe-GACs (a: virgin GAC; b: 4.22%; c: 9.29%; and d: 12.62%) Figure 9. XRD spectra of Fe-GACs Figure 10. Raw XRD spectrum for Norit RX3 EXTRA Fe-GAC with post-treatment (Iron content 7.53%) Figure 11. Impact of ph on arsenic adsorption on Darco Fe-GAC with 4.22% iron Figure 12. Impact of adsorption time on arsenic adsorption for Fe-GACs with different contents Figure 13. As(V) adsorption isotherm curves for Fe-GACs with different iron contents Figure 14. As(III) adsorption isotherm curves for Fe-GACs with different iron contents Figure 15. Maximum adsorption capacities of Fe-GACs for As(III) and As(V) Figure 16. BET surface area of Fe-GACs with different iron contents Figure 17. Comparison of calculated As(V) maximum adsorption capacity with experimental value Figure 18. As(V) maximum adsorption capacity vs. S Fe Figure 19. Distribution of surface area of Fe-GACs with pore size Figure 20. Adsorption isotherm curves for As(V) spiked groundwater (lines are Langmuir model with the same b values) Figure 21. Breakthrough curves of Fe-GACs with EBCT 600 s Figure 22. Relationship between iron content of Fe-GACs and BV treated/utilization rate at breakthrough iv

6 ACKNOWLEDGMENTS Stipend support and supplemental funding from the North Dakota Water Resources Research Institute (ND WRRI) are thankfully acknowledged. v

7 ABSTRACT Granular activated carbon (GAC) was impregnated with iron through a new multi-step procedure using ferrous chloride as the precursor for removing arsenic from drinking water. Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) analysis demonstrated that the impregnated iron was distributed evenly on the internal surface of the GAC. Impregnated iron formed nano-size particles, and existed in both crystalline (akaganeite) and amorphous iron forms. Iron-impregnated GACs (Fe-GACs) were treated with sodium hydroxide to stabilize iron in GAC and impregnated iron was found very stable at the common ph range in water treatments. The impact of the amount of impregnated iron in granular activated carbon (GAC) on arsenic adsorption capacity and kinetics was investigated at an arsenic concentration range that is typical for drinking water in this study. Results showed that the iron content of Fe-GAC reduces the arsenic adsorption rate and a five days adsorption time is necessary in the isotherm tests. The initial arsenic adsorption rate decreased with more impregnated iron, which is probably caused by the reduced pore size of Fe-GACs that is confirmed in the BET analysis. Langmuir model regression through a new approach disclosed that the amounts of impregnated iron affect arsenic adsorption capacity significantly. The maximum adsorption capacity (q m ) increased greatly with the iron content for both As(V) and As(III). When iron content is higher than 13.59%, more iron contributes less to arsenic adsorption capacity or even exhibits negative impact. Interestedly, the iron content has no impact on arsenic adsorption affinity. The adsorption affinity of Fe-GAC is L/μg for As(V) and L/μg for As(III). The specific surface area of impregnated iron was estimated according to the BET instrumental analysis of Fe-GACs. The surface area of impregnated iron shows the same trend as the arsenic adsorption capacity obtained in isotherm tests. At the iron content of 13.59%, impregnated iron possesses the highest specific surface area of 473 m 2 /g. Column tests showed that Fe-GACs can remove arsenic below the maximum contaminant level (10 µg/l) of arsenic in drinking water. The performance of Fe-GAC at breakthrough (10 µg/l) was enhanced significantly with increase of the amounts of impregnated iron. The number of bed volume (BV) treated at breakthrough increased from 140 to 1000 when the iron content of Fe-GACs increased from 4.56% to 13.59%. With further increase of iron content from 13.59% to 28.90%, BV treated at breakthrough slightly reduced from 1000 to 850. The utilization rate of Fe-GACs at breakthrough varied between 18.2% and 35.6%. The slope of breakthrough curves became smaller with increase of iron content, which implies that arsenic intraparticle diffusion rate in Fe- GACs decreased with iron content. Longer EBCT is required to achieve better use of Fe-GAC with high iron contents due to slow arsenic intraparticle diffusion. 1

8 INTRODUCTION Arsenic is a ubiquitous element in the environment and can be found in rocks, waters, and soils. It enters drinking water supplies from natural deposits in the earth as well as from anthropogenic activities, such as agricultural and industrial practices (1). Arsenic can cause serious health problems, such as cancers of bladder, lung, skin, kidney, liver, and prostate (2, 3). The USEPA has enforced a new arsenic standard for drinking water of 10 µg/l to protect people from the effects of long-term chronic exposure to arsenic since 2006 (4). Iron impregnated granular activated carbon (Fe-GAC) can remove arsenic from the aqueous phase effectively (5-12). Impregnation eliminates the disadvantages of using iron-based materials alone, such as poor mechanic strength, small specific surface area, and requirement for follow up filtration (6, 12). The primary advantage of GAC as the impregnation support is its huge internal surface area and advanced pore structure (13). When iron is evenly distributed inside GAC, the huge internal surface area of GAC can be largely used for iron impregnation and impregnated iron exhibits great adsorption capacity (11, 12). Since impregnated iron is recognized the active phase that enhances arsenic adsorption capacity of GAC, more iron was preferred in iron impregnation (5, 8). However, several studies showed that arsenic adsorption capacity did not increase proportionally to the increase of iron at high iron content (6, 12, 14). The percentage of arsenic adsorption was found to decrease with iron content when iron content was above 6% (6). Chang et al. (12) conducted a further study and used a more direct indicator, arsenic adsorption capacity, to demonstrate the relationship between iron content and adsorption capacity. They found that the maximum arsenic adsorption capacity was reached at the iron content of 4.22% and the iron impregnation method also affects the iron content that shows the maximum arsenic adsorption capacity. Even though both BET surface area and pore volume were reported to decrease with iron impregnation (6, 15-17), a direct link between arsenic adsorption capacity and BET surface/pore structure of Fe-GAC is still missing. It is still a speculation that the reduction of total surface area and blockage of pores caused arsenic adsorption capacity to reduce at high iron content. Further research is needed to systematically investigate how iron content affects the arsenic adsorption capacity. Adsorption kinetics is as important as adsorption capacity because it depicts the uptake rate of adsorbate and controls the equilibrium time of the adsorption process (18, 19). A few studies investigated arsenic adsorption kinetics on iron impregnated activated carbon and different kinetic models, including parabolic diffusion model, pseudo-first-order model, and pseudo-second order model, were used to interpret the arsenic adsorption (8, 17, 20). Because impregnated iron changes the GAC pore structure, it most likely will affect arsenic diffusion rate. Although a number of kinetic studies have investigated arsenic adsorption on Fe-GAC, little information is available on the effect of the amount of impregnated iron on arsenic adsorption kinetics. The objectives of this study were to: (1) develop a method to impregnate GAC with a high amount of iron that is stably and evenly distributed in GAC; (2) characterize Fe-GAC to determine the amount, distribution, morphology, and species of impregnated iron; (3) investigate impacts of the amount of impregnated iron on arsenic adsorption capacity and efficiency; (4) As (III) and As(V) adsorption characteristics of Fe-GACs with different iron contents at a low arsenic concentration range; and (5) the impact of iron content on the pore structure and surface area of Fe-GACs. 2

9 Materials MATERIALS AND METHODS Lignite-based GAC Darco (American Norit Co. Inc) was thoroughly washed using de-ionized (DI) water to clean impurities and powder, dried at 105 C overnight, and stored in desiccators for iron impregnation. Standardized As(III) (NaAsO 2 ) 0.1N solution (Alfa Aesar Co.) was used for preparation of As(III) solution. Sodium arsenate (Na 2 HAsO 4 ) was used for preparation of As(V) solution. Ultra Scientific certified standard As(V) solution (1000 mg/l in 2% nitric acid) was used for preparation of calibration standards in arsenic instrumental analysis. All other chemicals used in this study were of reagent grade and used as received, including ferrous chloride, sodium arsenate, hydrochloric acid, nitric acid, sodium hydroxide, and sodium bicarbonate. Groundwater samples were taken from a farmer s well at the former arsenic trioxide Superfund site (Site ID # ) near the City of Wyndmere, North Dakota (21). DO, temperature, specific conductivity, and ph were measured on-site; other parameters, total phosphate, TOC, turbidity, and arsenic, were determined in laboratory. Groundwater samples were stored at 4 C. Preparation of Fe-GAC The multi-step iron impregnation procedure developed by Chang et al. (12) was modified in this study by the addition of a stabilization step using sodium hydroxide. This sodium hydroxide post-treatment converts impregnated iron to ferric hydroxide which was identified later as akaganeite in X-Ray Diffraction (XRD) analysis. This stabilization step improves iron impregnation efficiency by reducing the lost of impregnated iron in the impregnation process, enhances the stability of impregnated iron, and increases arsenic adsorption capacity by generation of favored iron species - hydrous ferric hydroxide that has poor crystalline structure and large surface areas (22). The modified iron impregnation procedure is: (1) GAC reacts with ferrous chloride solution on a rotating shaker for 24 hr; (2) GAC particles are separated, and dried in a convective oven for 10 hr; (3) Fe-GAC is treated with 1 N NaOH solution for 10 hr; and (4) Fe-GAC is washed with DI water. The above process is repeated to produce Fe-GACs with desired iron amount. In this study, Fe-GACs have been prepared with 11 different iron contents ranging from 4.56% to 28.90%. SEM analyses indicated that iron impregnated through this method is evenly distributed inside GAC. Characterization of Fe-GAC Fe-GACs were evaluated based on four indicators: amount of impregnated iron, distribution of impregnated iron, stability of impregnated iron, and arsenate adsorption capacity of Fe-GAC. Iron content (defined in (Eq. (1)) was determined through an acid extraction method. Distribution and morphology of impregnated iron were measured by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS). Iron species were identified by X-ray diffraction (XRD). According to the objectives of this study, the pore structure and surface chemistry of Fe-GACs were not investigated. (1) 3

10 Measurement of iron content Three different methods, incineration (23), incineration and digestion (8), and acid extraction (24), were evaluated for the measurement of the amount of impregnated iron in GAC (Appendix III). The acid extraction method was selected and modified for this study. Three hundred milligrams of Fe-GAC was added into a 40-mL vial containing 30 ml 1:1 HCl solution and shaken for 10 h. Then, the vial was placed in a water bath at C for 4 h. The Fe-GAC was separated from the solution using GF/C filter paper. The iron concentration in the filtrate was analyzed using the phenanthroline method with a Hach spectrophotometer DR5000. Measurement of the distribution, morphology, and species of impregnated iron Fe-GACs were fractured to expose the internal structure for the examination of the distribution and morphology of impregnated iron using a JEOL JSM-6490LV scanning electron microscope equipped with a Thermo Energy Dispersive X-ray System. To avoid the destruction of the surface characteristics of Fe-GACs, no further polishing action was conducted on exposed cross-sections. XRD analysis was carried out to determine iron species in GACs using a Bruker AXS' D8 Discover diffractometer in Bragg-Brentano geometry, using Cu Kα radiation with a wavelength of nm. The powdered samples of Fe-GACs were scanned from 5 to 85, using a step size of 0.02 and a run time of 1 sec/step. Iron Desorption Test This test was conducted to evaluate the stability of impregnated iron. Five hundred milligrams of Fe-GAC was added to 40 ml DI-water and shaken 48 h. Fe-GACs were separated through filtration using GF/C filter paper. The iron concentration in the filtrate was analyzed using the same method described in Section Desorption of impregnated iron was evaluated according to the mass of dissolved iron in DI-water. Adsorption Kinetic Test One hundred milligrams of Fe-GAC were added into a number of 40-mL vials containing 35 ml 3mg/L As(V) solution with ph controlled at 7.0 using a 0.05N NaHCO 3 buffer solution. The bicarbonate buffer solution was found to have no impact on arsenic adsorption using iron based materials (25, 26). These vials were placed on the rotating shaker at 30 rpm and 25 C. A vial was taken from the shaker to determine the arsenic concentration in the aqueous phase. The filtration process used 0.45 µm membrane filter papers. Adsorption kinetic tests were carried out for Fe-GAC with iron contents of %. Adsorption Isotherm Test The arsenic adsorption capacities of Fe-GACs were evaluated in batch experiments. One hundred milligrams of Fe-GAC was added into 40-mL vials containing 35 ml arsenic solutions. The initial concentrations of the arsenic solution for isotherm tests varied between 200 µg/l and 3000 µg/l, which was to achieve the desired arsenic equilibrium concentration of < 250 µg/l to simulate the range of typical water treatment. To prevent the oxidation of As(III), nitrogen purged DI-water was used in preparation of As(III) solutions. The ph of adsorption tests was controlled at 7.0 using a 0.05 N NaHCO 3 buffer solution. After 5 days of shaking at 30 rpm and 25 C, Fe-GACs were separated from the solution using 0.45 µm membrane filter papers. The 4

11 filtrate was preserved by 6 N HNO 3 and stored at 4 C for arsenic analysis. The same procedure was used for the adsorption isotherm tests on groundwater samples. Fe-GAC Surface Area and Pore Volume Measurement The liquid nitrogen adsorption-desorption isotherms of Fe-GACs with different iron contents at 77 K were measured using Micro-meritics ASAP 2020 to characterize the surface area and the porous structure of Fe-GACs. Prior to measurement, Fe-GACs were degassed at 105 C under nitrogen flow for 12 hr. BET model was used for the calculation of surface area of Fe-GACs. Column Test Borosilicate glass columns with an inner diameter of 7 mm were used in breakthrough tests. The mean size of Fe-GAC is 0.5 mm so that the ratio of diameter of column to particle size of adsorbent is around 14, which is smaller than commonly recommended value of 20 to avoid channeling (19). Because the main objective of column tests in this study is to evaluate the impact of iron content of Fe-GACs on arsenic breakthrough rather than a pure design purpose, the small size of column is acceptable. In addition, special measures were taken to minimize channeling. Fe-GACs were soaked in DI-water for 48 hours to degas prior to addition to columns. To avoid entrainment of air during column preparation, wetted Fe-GAC was transferred to columns filled with de-ionized (DI) water. DI-water was used for initialization of Fe-GAC packed column for flow-rate adjustment and compaction of Fe-GAC bed. Then, influent was switched to groundwater to start tests. At two ends of Fe-GAC bed, glass wool was added to avoid loss of Fe-GACs during tests. All tubes used were Tygon tubing and fittings were HDPE or PVC to minimize the possible arsenic adsorption. Arsenic solution was continuously fed into the column using a Cole-Parmer Master peristaltic pump in an up-flow pattern to minimize channeling as well. In the full-scale column parallel tests, it was found that EBCTs of 6.3 min and 3 min with granular ferric hydroxides (GFH) resulted in nearly identical arsenic breakthrough curves. Samples were collected from the sampling port at the up-end of column at predetermined time and filtered through 0.45 membrane syringe filter. Filtrate was saved at 4 C for arsenic analysis. The flow-rate was monitored daily. Analytical Methods Arsenic concentration in water samples was determined using a Perkin Elmer 4110ZL Graphite Furnace Atomic Absorption Spectrometer (GFAAS) at nm in accordance with the EPA method Rev.3.0 (27). A calibration curve with arsenic concentrations of 0, 2, 5, 15, 30, and 60 µg/l was prepared using an As(V) standard solution (1000 µg/ml in 2% nitric acid, Ultra Scientific certified). A linear response with correlation coefficient (R 2 ) higher than was achieved. This method had an arsenic detection limit of 1 µg/l. For the analytical quality control, an instrument performance check (IPC) solution (30 µg/l) and calibration blank were analyzed after every 10 samples. The GFAAS was programmed to recalibrate the arsenic analytical method if a 10% variation of IPC was observed. Additionally, spiked experimental blanks/samples were adopted as quality control measures and the recovery rate was %. The iron concentration was analyzed through the phenanthroline method with a Hach spectrophotometer DR

12 RESULTS AND DISCUSSION Preparation of Fe-GAC It is difficult to achieve high amounts, even distribution, and stability for impregnated iron inside GAC in one step. This new synthesizing method is a multi-step procedure in which a small amount of iron is impregnated in GAC to achieve full penetration and even distribution in each step. Then, by repeating this process, high amounts of iron can be achieved as well as an even iron distribution inside GAC. Ferrous was used as the precursor because it is soluble at a wide range of ph and can diffuse deep into the internal pores of GAC (6). However, ferrous impregnated in GAC may not be stable because it can re-dissolve into solution. Therefore, in the synthesis process, the drying step was purposely extended to 10 h to stabilize iron inside GAC by the conversion of ferrous to ferric using oxygen in air in the convective oven. In this step, most ferrous chloride is expected to be oxidized to ferric chloride, ferric oxide, and ferric hydroxide. Two GACs, Darco and Norit RX3 EXTRA, were used for iron impregnation to examine the capability of this synthesizing method. As shown in Figure 1, iron contents were 2.03% for Darco and 0.77% for Norit RX3 EXTRA after the first 24 h treatment. Iron contents increased almost linearly with repetitions, 1.03% per repetition for Darco and 0.74% per repetition for Norit RX3 EXTRA. If all pores of GAC are assumed to be filled with 0.5 M ferrous solution ideally in each repetition, according to the total pore volumes of GACs, the impregnated iron is 26.6 mg/g for Darco and 19.4 mg/g for Norit RX3 EXTRA per repetition. Therefore, the ideal increment of iron content in each repetition will be 2.60% for Darco and 1.90% for Norit RX3 EXTRA. The experimental value of iron increment per repetition is about 40% of calculated values for both GACs. Most likely some of impregnated iron re-dissolved into the liquid phase in the following repetitions to cause the difference between experimental values and calculated values. This synthesizing method is easy to conduct and the iron content can simply be controlled by the number of repetitions. After 10 repetitions, 12.62% and 8.52% iron were impregnated in Darco and Norit RX3 EXTRA, respectively. More iron may be impregnated with more repetitions. Stability of Impregnated Iron Iron desorption tests were carried out firstly to examine the stability of impregnated iron and results indicated that without post-treatment some of impregnated iron was unstable. Depending on iron content of Fe-GACs, 6-46% of impregnated iron were lost in desorption tests. EDS analysis revealed that large amounts of chlorine in Fe-GACs (Figure 2.a), which implied the existence of ferric or ferrous chlorides. Ferric and ferrous chlorides may dissolve into solution because ferric chloride and ferrous chloride have solubility of 92 g/100 ml and 68.5 g/100 ml in water at 20 C, respectively. 6

13 Iron Content (%) 15 Norit RX3 EXTRA Darco y = x R² = y = x R² = Repetitions Figure 1. Increment of iron content with repetitions in the synthesis process. A post-treatment was designed to further stabilize impregnated iron through conversion of iron chlorides to ferric hydroxide which is sparingly soluble (Table 1). After post-treatment, desorption tests were conducted again and results showed that the post-treatment successfully stabilized iron in Fe-GACs. EDS analysis, conducted on post-treated Fe-GACs, indicated that the amounts of chlorine reduced significantly (Figure 2.b). The stability of impregnated iron was also tested on Fe-GAC with 4.22% iron in the adsorption test with ph variation of As shown in Figure 3, impregnated iron dissolved when ph was below 4. At ph 2.08, dissolved iron concentration was 25 mg/l, which is equivalent to 21% of impregnated iron. Impregnated iron was very stable when ph was above 4. Because ph is usually neutral in water resources, no concern is necessary for the iron stability of Fe-GACs. The iron stability was also monitored in all adsorption tests using ICP and the dissolution of impregnated iron was negligible. a 7

14 Dissolved Fe concentration (mg/l) b Figure 2. EDS spectra, a: Darco Fe-GAC (12.68%) without post-treatment, b: Darco Fe-GAC (9.29%) with post-treatment. Table 1. Dissociation Constants of Iron Hydroxides Iron Species pk1 pk2 pk3 pka Fe(OH) 3 (Amorphous) From Gunter Faure, Fe(OH) ph at equilibrium Figure 3. Stability of impregnated iron on Darco Fe-GAC with 4.22% iron. 8

15 Iron relative percentage % Characterization of Fe-GAC Distribution of impregnated iron Darco Fe-GACs, with iron contents of 4.22%, 9.29% and 12.62%, were analyzed using SEM/EDS to investigate the iron distribution inside GAC. Figure 4 shows the relative weight percentages of iron in the four zones from the edge to the center on the cross-section of Fe-GACs (Figure 5), which indicated that iron was evenly distributed on the cross-section of GACs rather than concentrated on the edge. The variation of the iron relative weight percentages across the cross-section of Fe-GAC is likely caused by the heterogeneity of pore sizes and morphologies of GAC itself. The even iron distribution was found in Norit RX3 EXTRA Fe- GAC with iron content of 8.52 % as well. As shown in Figure 4, relative weight percentages of iron are different between SEM/EDS analysis and the results from acid extraction method described in Section With the acid extraction method as reference, SEM/EDS analysis generated underestimate at low iron content (4.22%), overestimate at high iron content (12.62%), and matched estimate with medium iron content (9.22%). EDS detects elements on the surface of samples with certain penetration depth. For Fe-GAC with low iron content, it is expected that iron distributed evenly on GAC internal surfaces in both macro-pores and micro-pores. Because most internal surface areas are contained in micro-pores and crosscutting of the Fe-GAC samples will not expose most micro-pores, EDS is not able to detect unexposed iron in micro-pores and resulted in underestimated iron contents. At high iron contents, iron crystals or nano-scale iron deposits may form in macro-pores. Exposure of large amount of iron in macro-pores could be the reason for overestimation of iron content from EDS analysis. (More EDS analyses are included in Appendix VI and VII) % 9.29% 12.62% Zone Figure 4. Iron distribution on the cross-section of Darco Fe-GACs. 9

16 In addition to EDS analysis, mapping of iron distribution on the cross-section of Darco Fe-GAC with 9.29% iron shows that impregnated iron covered the entire cross-section of Fe-GAC and higher iron content was observed in the macro-pores (cracks) of Fe-GAC (Figure 6) Figure 5. EDS analysis on the cross-section of Darco Fe-GAC with 4.22% iron. a b Figure 6. Mapping of iron distribution on cross-section of Darco Fe-GAC with 9.29% iron. a:the cross-section of Fe-GAC; b: the mapping of iron. Morphology of impregnated iron Internal structure and morphology of Norit RX3 EXTRA Fe-GAC with 8.52% iron are shown in Figure 7. Iron formed rod-shape nano-size particles in macro-pores of GAC. In Figure 7b, iron rods were measured approximately 300 nm in length and 40 nm in diameter. Areas without iron rods in Figure 7a are believed to be areas with micro- and meso-pores which can be seen from lower left part of Figure 7b. Due to the constraint of pore size, impregnated iron 10

17 cannot form rod-shape nano-particles in these areas; however, iron is believed to exist in a smaller scale. Figure 8 shows morphologies of virgin and iron impregnated Darco GAC. Instead of rod-shape nano-particles, irregular shapes of iron deposited in Darco Fe-GAC with low iron content (Figures 8b and 8c) while nano-iron crystals were observed in Fe-GAC with high iron content (Figure 8d). a b Figure 7. SEM images of Norit RX3 EXTRA Fe-GAC with 8.52% iron. b a b c d Figure 8. SEM images of Darco Fe-GACs (a: virgin GAC; b: 4.22%; c: 9.29%; and d: 12.62%). 11

18 Arbitrary Intensity (cps) Iron species Both post-treated and non-post-treated Fe-GACs were pulverized for XRD analyses to identify iron species. The major peaks of processed XRD spectra (Figure 9) matched well with the crystalline iron species akaganeite (β-feooh); while, the raw XRD spectrum (Figure 10) implies that significant amounts of amorphous iron existed in Fe-GAC as well. The species of impregnated iron are affected by impregnation methods and synthesizing conditions (22, 28). Jang et al. (22) reported that impregnated iron formed akaganeite at synthesizing temperature 80 C and amorphous hydrous ferric oxide (HFO) at 60 C. However, Gu et al. (6) found that impregnated iron in GAC, synthesized at 80 C, were all in amorphous state. Results of this study found that impregnated iron existed as both crystalline and amorphous species in GAC. Norit RX3 EXTRA Fe-GAC Darco Fe-GAC Norit RX3 EXTRA Fe-GAC without post-treatment Darco Fe-GAC without post-treatment θ (Degrees) ph Impact on Arsenic Adsorption Figure 9. XRD spectra of Fe-GACs. ph is one of the most important factors affecting arsenate adsorption in the liquid phase (6,14,15, 22,29), because the arsenate species and the surface charge of Fe-GACs in liquid phase depend on ph. The impact of ph on arsenate removal was studied with ph values ranging from 2 to 11 using 100 mg Darco Fe-GAC with 4.22% iron and 35 ml 3 mg/l arsenate solution. As shown in Figure 11, arsenic removal rate maintained close to 100% at ph 2-6 and declined slightly at ph When ph was above 7.0, arsenic removal rate declined quickly. To minimize the impact of ph variation on isotherm tests as well as to keep experimental conditions close to natural waters, the ph in arsenate isotherm tests was controlled at 7.0 using bicarbonate buffer solution. 12

19 Intensity (cps) θ (degree) Figure 10. Raw XRD spectrum for Norit RX3 EXTRA Fe-GAC with post-treatment (Iron content 7.53%). To evaluate the impact of ph on arsenate adsorption, major arsenate species in different ph ranges are included in Figure 11. Arsenate species in aqueous phase exist mainly as H 3 AsO 4 at ph less than 2.2 (pk 1 ), H 2 AsO 4 - at ph between 2.2 and 6.98 (pk 2 ), HAsO 4-2 at ph between 6.98 and 11.5 (pk 3 ), and AsO 4-3 at ph above ph zpc is the ph at which adsorbents have a net zero surface charge (30, 31). The ph zpc for Fe-GAC was between 8.2 and 8.7 (15). As shown in Figure 11, Fe-GACs becomes more positively charged when ph is less than 8.3 (ph zpc ) and more negatively charged when ph is above 8.3. As ph increased, the attractive force between Fe-GAC and arsenate became less and changed to repulsive force when ph above 8.3. The change of the electrostatic force between arsenate species and Fe-GAC explained the ph impact on arsenate adsorption. 13

20 Arsenate Removal Rate Charge 100% 80% Fe-GAC 60% 40% H 3 AsO 4 0 H 2 AsO 4-1 HAsO 4-2 AsO 4-3 Arsenate 20% 0% Equilibrium ph Figure 11. Impact of ph on arsenic adsorption on Darco Fe-GAC with 4.22% iron. Arsenic Adsorption Equilibrium Time Adsorption rate and time required for reaching equilibrium are a function of several factors, including activated carbon particle size, pore structure and adsorbate characteristics (30). It may take a few hours to reach equilibrium for adsorption of contaminants with small molecular sizes on powered or pulverized activated carbons, while few months may be necessary for adsorption of larger molecules on GAC to reach true equilibrium (30, 32). Impregnation of iron on GAC not only changes the GAC surface characteristics in favor of arsenic adsorption, but also alters the GAC pore sizes and structures. Reduction of pore size and pore volume of activated carbon due to iron impregnation has been documented by several research groups (6, 14, 23). Volume reduction was found to be linearly related to the amount of iron content and it was speculated that iron particles may block or partially block some of the pores making them not available for arsenic removal or slowing the adsorption process. Because intraparticle diffusion rate and overall arsenic adsorption kinetics are likely affected by iron impregnation, longer equilibrium time may be required for Fe-GAC with high iron contents. Therefore, it is necessary to find an adequate adsorption duration for reaching adsorption equilibrium so that the impact of iron content on arsenic adsorption capacity can be properly evaluated. Fe-GACs with five different iron contents ranging from 5.57% to 28.90% were used to study As(V) removal versus adsorption time. Experiments were run for 15 days with samples taken at predetermined intervals for arsenic analysis. Amount of arsenic adsorbed at each sampling time was normalized by the adsorption capacity at 15 days and results are presented as percentage of adsorption capacity in Figure 12. Arsenic removal occurred rapidly in 14

21 Normaalized adsorption capacity the first 24 hours, reaching 80% or higher of the maximum capacities for all Fe-GACs. Much slower removal rates were observed afterward. It took 48 hours to reach 90% of total capacities and 120 hours to achieve 95%. It is also observed that adsorption on each Fe-GAC proceeded at different rates and the difference is most noticeable during first 24 hours (see Figure 12 insert). In general, Fe-GACs with higher iron contents exhibits slower arsenic removal rate. This may be explained by slower arsenic intraparticle diffusion rate due to the GAC pore size reduction from iron impregnation. Figure 12 shows that the percentage of total adsorption capacity realized is getting close to each other for Fe-GACs with different iron contents after 24 hours. It has to be noted that a small percentage difference in total adsorption capacity for Fe-GACs with high iron contents may represent significant capacity for arsenic removal, because the total (maximum) arsenic adsorption capacities for different iron contents (presented later in this paper) are different, Based on the results of this study and practical consideration, a 5-day adsorption time was selected for arsenic adsorption isotherm tests in this study. 120% 80% 40% 0% Adsorption time (hr) 5.57% 8.19% 18.01% 24.73% 28.90% Figure 12. Impact of adsorption time on arsenic adsorption for Fe-GACs with different contents. As(III) and As(V) Adsorption Isotherms Fe-GACs with eleven iron contents ranging from 1.64% to 28.90% were tested to investigate the impact of iron contents on arsenic adsorption properties at a low concentration range (<250 μg/l) which is typical in contaminated drinking water sources. Contaminated water sources may contain both As(III) and As(V). Therefore, both As(III) and As(V) were tested in isotherm tests. The ph of adsorption isotherm tests was controlled in a narrow range to minimize its impact. As shown in Figures 13 and 14, As(V) and As(III) adsorption capacities on Fe-GACs were enhanced significantly with increase of iron content. Virgin GAC was tested as well, while the result is not presented in Figure 13 due to its negligible arsenic adsorption capacity. 15

22 q (µg/g) Langmuir model (Eq. 2) was chosen to interpret the experimental data because the nature of arsenic adsorption on Fe-GAC is site specific adsorption (33, 34). In order to avoid biased estimates from linearization of Langmuir model, a non-linear least squares regression procedure was developed using Statistical Analysis System (SAS) to estimate the parameters in Langmuir model (12). Langmuir parameters obtained from the best fits of As(V) and As(III) experimental data are shown in Table 2 and model simulations are shown as solid lines in Figures 13 and 14. As shown in Table 1 and Figures 13 and 14, Langmuir model fits As(V) and As(III) adsorption equilibrium data well with correlation coefficient (R 2 ) of 0.96 or better. (2) Ce (µg/l) Langmuir model fit 1.64% 4.56% 8.19% 5.57% 9.96% 11.49% 13.59% 15.03% 18.01% 24.73% 28.90% Figure 13. As(V) adsorption isotherm curves for Fe-GACs with different iron contents. 16

23 q (µg/g) 1,600 1,400 1,200 1, Ce (µg/l) Langmuir model fit 1.64% 5.57% 8.19% 11.49% 13.59% 15.03% 18.01% 24.73% 28.90% Figure 14. As(III) adsorption isotherm curves for Fe-GACs with different iron contents. Table 2. Langmuir Model Parameters for As(III) and As(V) Adsorption on Fe-GACs Iron content (%) As(V) As(III) ph q m (µg/g) b (L/µg) R 2 ph q m (µg/g) b (L/µg) R / / / / / / / / The symbol of / indicates that the experimental data is not available. 17

24 Adsorption Affinity of Arsenic on Fe-GACs The parameter b in the Langmuir model represents the affinity of an adsorbate to an adsorbent. From the reaction kinetics aspect, the parameter b in Langmuir model is the ratio of the adsorption rate constant to desorption rate constant (32, 35). According to the reaction thermodynamics, the parameter b is related to the net enthalpy of adsorption that reflects the strength of binding of the adsorbate to the adsorbent (30, 36, 37). Therefore, higher b value indicates higher adsorption affinity that means most adsorption capacity can be utilized at low equilibrium concentration (12). As shown in Tables 2, the values of parameter b vary in a narrow range for Fe-GACs with different iron contents, except for iron content of 1.64%. The average b value and the standard deviation are L/ g for As(V) and L/ g for As(III), respectively. Because Fe-GACs in this study were synthesized by the same procedure, the surface characteristics of impregnated iron are expected to be the same. The Langmuir constant b is, therefore, expected to be the same for Fe-GACs with different iron contents. Based on the above consideration, a hypothesis was postulated that the amounts of impregnated iron of Fe-GACs only affect the arsenic maximum adsorption capacities but not the affinity for arsenic. To determine whether the isotherm data support this hypothesis, the non-linear regression process was repeated with the same b value for all Fe-GACs with different iron contents. The average b values for As(V) and As(III) adsorptions were used as common values in the nonlinear regression. Results of this analysis show that Langmuir model with the same parameter b value can still well fit the adsorption isotherms for both As(III) and As(V) (Table 3). This verifies that Fe-GACs with different iron contents possess the same affinity for arsenic. Table 3. Re-estimated q m in Langmuir Model with same b Values for As(III) and As(V) Adsorption on Fe-GACs Iron content (%) As(V) As(III) q m (µg/g) b (L/µg) R 2 q m (µg/g) b (L/µg) R

25 It is noticed that Fe-GAC with iron content of 1.64% has the poorest Langmuir model fits for both As(III) and As(V) in Table 3, which implies that Fe-GAC with iron content of 1.64%, as a whole adsorbent, presents very different surface characteristics from others. This is probably because the iron content of 1.64% is so low that only a small fraction of the surface of GAC is covered by iron. The overall surface characteristics of Fe-GAC with iron content of 1.64% are mainly determined by the bare surface of GAC that has very low affinity for arsenic adsorption. When much higher affinity is assigned to Fe-GAC with iron content of 1.64%, poor Langmuir model fit is anticipated. Parameter q m of Langmuir model is commonly used to evaluate and compare arsenic adsorption of Fe-GACs produced from different GACs and impregnation methods in literature (6, 38-41). The parameter b, however, draws less attention even though it has significant influence on adsorption capacity at low equilibrium concentrations. The adsorption affinity of Fe-GACs synthesized in this study, L/µg for As(V) and L/µg for As(III), are comparable or higher than the values reported in literature (6, 8, 16), indicating that Fe-GACs synthesized in this study are suitable for applications on arsenic contaminated drinking water. High affinity of arsenic to Fe-GACs observed in this study probably is attributed to two reasons: the favored surface characteristics of impregnated iron and low arsenic concentration in isotherm tests. XRD analysis on Fe-GAC samples showed that the dominating iron species on GAC surfaces is akageneite, which is considered one of the most favored hydrous ferric oxide species for arsenic removal (33). Authors identified an interesting observation in literature that that the statistical regression analysis on Langmuir model tends to return high q m and low b when isotherm tests were conducted at high equilibrium concentration, while low q m and high b at low equilibrium concentrations (8, 29, 42). This study intends to evaluate the adsorption properties of Fe-GACs at conditions close to real cases, usually low arsenic concentration. Therefore, the isotherm test conditions may contribute to high b values in Langmuir model analysis. Impact of Iron Content of Fe-GACs on Arsenic Adsorption Capacities The arsenic maximum adsorption capacity, q m, was used to evaluate the impact of the iron content of Fe-GACs on the arsenic adsorption capacity because q m is independent of the arsenic equilibrium concentration. As shown in Table 3 and Figure 15, q m for As(V) increased from 265 µg/g to 1996 µg/g when impregnated iron increased from 1.64% to 13.59%. Then, the q m gradually decreased to 1445 µg/g with impregnated iron increasing to 28.90%. For As(III), q m quickly increased from 454 µg/g to 2116 µg/g when impregnated iron increased from 1.64% to 13.59%. Further increase of iron contents resulted in slower increase of q m, which reached 2456 µg/g at iron content of 24.73%. As shown in Figure 15, when iron content is below 13.59%, the q m for both As(III) and As(V) increased linearly with iron content and are at approximately same values. This is probably because the adsorption sites on Fe-GACs are accessible to both As(III) and As(V), and the number of adsorption sites is proportional to the amount of impregnated iron. When iron content is higher than 13.59%, adsorption capacity is compromised by partial blockage of pore structures or covering up adsorption sites (detailed explanation in the next section). Similar observations were documented in the previous studies (6, 23). It is speculated that the accessibility of adsorption sites is different for As(III) and As(V) due to the different chemical properties and physical size of these two arsenic species on Fe-GACs with high iron contents. 19

26 q m (µg/g) As(III) As(V) Iron content (%) Figure 15. Maximum adsorption capacities of Fe-GACs for As(III) and As(V) BET Surface Area and Pore Structures of Fe-GACs Pore size distribution and BET surface area of Fe-GACs were analyzed to investigate the impact of iron content on arsenic adsorption capacities. Fe-GACs with iron content of % were selected for the analysis of BET surface area and pore structure. Figure 16 shows the BET surface area of Fe-GAC (S Fe-GAC ) decreased from 554 m 2 /g to 308 m 2 /g as the iron content increased from 0% (virgin GAC) to 28.90%. The average deduction of S Fe-GAC per percent of impregnated iron is 6 m 2 /g %Fe, less than m 2 /g %Fe reported in literature (6, 16). The smaller deduction rate of S Fe-GAC implies that impregnated iron is evenly distributed on the internal surface area of GAC. S Fe-GAC is usually used to explain the arsenic adsorption capacity in literature because it is readily reported from the instrumental analysis. However, S Fe-GAC contains the surface area of both impregnated iron and GAC and cannot be directly related to arsenic adsorption capacity because the surface area of impregnated iron (S Fe ) determines arsenic adsorption capacity. 20

27 BET Surface area of Fe-GAC (m 2 /g) Iron content (%) Figure 16. BET surface area of Fe-GACs with different iron contents. S Fe can be estimated from S Fe-GAC and their relationship changes with iron contents. When a small amount of iron is impregnated in GAC, iron (in this case akaganeite) generate active surface for arsenic adsorption. At the same time, internal surface area of GAC is covered and the pore volume is occupied by impregnated iron, which causes an overall deduction of S Fe-GAC. At low iron contents, S Fe increases with increase of iron content, while S Fe-GAC decreases. When the internal surface of GAC is fully covered by iron and a maximum S Fe is reached, S Fe is equal to S Fe-GAC and further increase of iron will result in decrease of both S Fe-GAC and S Fe. Because adsorption capacity of arsenic is proportional to S Fe, q m will increase with increase of S Fe and decrease with decrease of S Fe. Based on the above considerations and the highest q m obtained at iron content of 13.59%, it is assumed that the impregnated iron fully covers the internal surface area of GAC and reached a maximum S Fe at iron content of 13.59%. Therefore, in the range of %, S Fe increases with iron content while S Fe-GAC decreases with iron content. In the range of %, both S Fe and S Fe-GAC decrease with iron content. With the assumption that impregnated iron fully covers the internal surface of GAC at 13.59%, S Fe was calculated for Fe-GACs with different iron contents. Then, the q m was calculated based on the iron content, S Fe, and arsenic specific adsorption density. Comparison of calculalted q m with experimental value in Figure 17 shows that the calculated As(V) q m agreed with the experimental data well, which verified the assumption that the internal surface of GAC is fully covered by impregnated iron at iron content of 13.59%. This assumption was confirmed by the same analysis on As(III) as well. 21

28 As (V) adsorption density (µg As/g Fe) q m (µg/g) Experimental value Calculated value Iron content (%) Figure 17. Comparison of calculated As(V) maximum adsorption capacity with experimental value Figure 18 depicts the relationship between the As(V) adsorption density, defined as ratio of q m to iron content of Fe-GAC), and S Fe. As show in Figure 18, As (V) adsorption density remains at approximately the same level before GAC is fully covered with impregnate iron (S Fe 473 m 2 /g at iron content of 13.59%). Then, as iron content keeps increasing (> 13.59%), arsenic adsorption density decreased with decreasing of S Fe because of the formation of multi-layer of impregnated iron Fe BET surface area (m 2 /g Fe-GAC) Figure 18. As(V) maximum adsorption capacity vs. S Fe 22

29 Cumulative surface area (m 2 /g) Figure 19 shows the distribution of S Fe-GAC with pore size, which indicates that more than 73% of S Fe-GAC was contributed from pores with size less than 2 nm in the tested range of pore size (< 40nm). This is consistent with the finding in the kinetic study that long adsorption time is necessary to reach the adsorption equilibrium Pore diameter (nm) virgin GAC 1.64% 5.57% 8.19% 13.59% 18.01% 24.73% 28.90% Figure 19. Distribution of surface area of Fe-GACs with pore size. The pore volume of Fe-GACs reduced linearly from cm 3 /g to cm 3 /g when iron content increased from 0% to 28.90% (Table 4), which implies that the density of impregnated iron are for Fe-GACs with different iron contents. This agrees with the finding in the arsenic isotherm tests that the surface characteristics of impregnated iron are the same for Fe- GACs. With iron impregnation, the average meso-pore size of Fe-GACs continuously reduced from 9.95 nm to 7.48 nm, which may explain why the initial arsenic adsorption rate decreased with more impregnated iron (Figure 12). Table 4. Texture Parameters for Fe-GACs and Virgin GAC Iron content (%) S Fe-GAC Pore Volume Average pore (m²/g) (cm³/g) diameter (nm) Virgin GAC